WATER RESOURCES RESEARCH, VOL. 39, NO. 4, 1084, doi:10.1029/2002WR001503, 2003
Influence of transitional volcanic strata on lateral diversion at
Yucca Mountain, Nevada
Lorraine E. Flint and Alan L. Flint
Water Resources Division, U.S. Geological Survey, Sacramento, California, USA
John S. Selker
Department of Bioresource Engineering, Oregon State University, Corvallis, Oregon, USA
Received 5 June 2002; revised 14 October 2002; accepted 15 January 2003; published 3 April 2003.
[1] Natural hydraulic barriers exist at Yucca Mountain, Nevada, a potential high-level
nuclear waste repository, that have been identified as possible lateral diversions for
reducing deep percolation through the waste storage area. Historical development of the
conceptual model of lateral diversion has been limited by available field data, but numerical
investigations presented the possibility of significant lateral diversion due to the presence of
a thin, porous rock layer, the Paintbrush nonwelded tuffs. Analytical analyses of the
influence of transitional changes in properties suggest that minimal lateral diversion is
likely at Yucca Mountain. Numerical models, to this point, have not accounted for the
gradual transition of properties or the existence of multiple layers that could inadvertently
influence the simulation of lateral diversion as an artifact of numerical model discretization.
Analyses were made of subsurface matric potential measurements, and comparisons were
made of surface infiltration estimates with deeper percolation flux calculations using
chloride-mass-balance calculations and simulations of measured temperature profiles.
These analyses suggest that insignificant lateral diversion has occurred above the repository
horizon and that water generally moves vertically through the Paintbrush nonwelded
INDEX TERMS: 1875 Hydrology: Unsaturated zone; 5104 Physical Properties of Rocks: Fracture and
tuffs.
flow; 1829 Hydrology: Groundwater hydrology; 5114 Physical Properties of Rocks: Permeability and porosity;
KEYWORDS: volcanic tuffs, lateral diversion, capillary barrier, Yucca Mountain
Citation: Flint, L. E., A. L. Flint, and J. S. Selker, Influence of transitional volcanic strata on lateral diversion at Yucca Mountain,
Nevada, Water Resour. Res., 39(4), 1084, doi:10.1029/2002WR001503, 2003.
1. Introduction
[2] Yucca Mountain, Nevada (Figure 1), is being studied
to determine whether it is suitable for storage of high-level
radioactive waste. The site was selected as a potential
repository partially because of the abundance of natural
barriers to the migration of water that could possibly interact
with and corrode the waste canisters [Roseboom, 1983].
One of the natural barriers identified early in the characterization was the layered, nonwelded volcanic tuffs that exist
between the ground surface and the potential repository
horizon, and the potential of these tuffs to laterally divert
water away from the waste storage area.
[3] The maintenance of a relatively dry subsurface is of
utmost concern in waste containment. This situation can
occur when no water penetrates the surface or when the
infiltrating water is diverted laterally by way of a sloping
permeability or capillary barrier, either natural or manmade. This paper presents an evaluation of the potential
for lateral diversion by natural barriers.
[4] Waste containment is often a relatively small scale
endeavor that is controlled by careful engineering. The
situation at Yucca Mountain, on the other hand, is large
scale, with the potential subsurface repository for high-level
Copyright 2003 by the American Geophysical Union.
0043-1397/03/2002WR001503
SBH
radioactive waste, extending over 4 km2, at a depth in the
unsaturated zone below ground surface of 30 m. The
initial investigations identified low precipitation (170 mm/
yr) [Hevesi et al., 1992] and natural geologic barriers (a
deep unsaturated zone (500 – 1000 m) composed of Tertiary
volcanic sequences of ash-flow and ash-fall tuffs with
alternating layers of welded and nonwelded tuffs [Buesch
et al., 1996]) as critical attributes of the site. The highly
fractured volcanic rocks surrounding the potential repository were expected to drain any percolating water quickly
[Roseboom, 1983]. In addition, it was assumed, on the basis
of early conceptual models, that water would be diverted
laterally via the thin, sloping, high-permeability, nonwelded
layer situated between the upper two thick, fractured, and
welded units, the Tiva Canyon and Topopah Spring Tuffs
[Montazer and Wilson, 1984; Flint et al., 2001a] (Figure 2).
[5] Natural hydraulic diversion mechanisms may exist on
large scales in deterministic depositional environments that
provide large-scale features with contrasting hydrologic
properties. The question posed here is whether an ashflow/ash-fall tuff environment, such as that at Yucca Mountain, provides those features, and on a scale pertinent to the
intent necessary at this site. The concept of a natural
capillary barrier is summarized by Montazer and Wilson
[1984] as a fine-grained layer overlying a coarse-grained
layer, where water cannot flow from the smaller pores of the
4-1
SBH
4-2
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
Figure 1. Location of study site, potential repository boundary, underground tunnels, and boreholes.
Figure 2. Cross-sectional schematic of major lithostratigraphic units at Yucca Mountain, and the
locations of the potential repository, Exploratory Studies Facility, and Cross Drift. Scale is approximate
with vertical scale 5 times the horizontal scale.
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
SBH
4-3
fine-grained Tpt. In contrast, if flow though the fine-grained
rock is predominantly through the fractures, as occurs at high
net-infiltration rates, then the opposite scenario would apply.
The high conductivity fractures in the Tpc overly a relatively
lower permeability porous matrix (the PTn has few fractures)
causing a permeability barrier at the base of the Tpc and the
top of the PTn. A capillary barrier exists between the
relatively lower permeability PTn and the higher-permeability fractures of the underlying Tpt (Figure 3).
[6] In fractured rock systems the amount of percolation
can determine whether the flow is fracture- or matrixdominated and hence whether the layer will act as a
capillary barrier or a permeability barrier. Because of the
variability in net infiltration at Yucca Mountain, both permeability and capillary barriers can be in effect at either the
base of the Tpc or the base of the PTn. These conditions are
further complicated by the degree of saturation of the matrix
and the size of the fracture apertures and whether the
fractures are open or filled.
[7] The purpose of this work is to evaluate evidence for
a lack of lateral diversion on any significant scale at Yucca
Mountain above the potential repository horizon. This will
be done by discussing (1) how the conceptual model of
subsurface flow that described the lateral diversion of
water at Yucca Mountain changed over the years on the
basis of an increase in available data and information, (2)
how field data constrained calculations and numerical
models describing lateral flow, (3) analytical calculations
describing capillary barrier mechanisms on the basis of
available properties from the site, and (4) data and interpretations representing various scales that support downward flow through the nonwelded layers and a lack of
significant lateral diversion.
Figure 3. Lithostratigraphic schematic of the nonwelded
rocks of the Paintbrush Group (PTn) and surrounding
welded rocks, and hydrogeologic unit descriptions.
fine-grained layer into the larger pores of the coarse-grained
layer until the height of water in the overlying layer exceeds
a critical height, equivalent to the difference in the capillary
rise (air-entry potential) of the two pore sizes. At Yucca
Mountain, these definitions are complicated by the assumed
flow mechanisms and must be carefully defined. Yucca
Mountain consists of a series of fine- and coarse-grained
rock layers (Figures 2 and 3), but for simplifying the
definitions we assumed that a fine-grained, low-permeability
rock (Tiva Canyon Tuff, Tpc) overlies a high-permeability,
coarse-grained rock (nonwelded rocks of the Paintbrush
Group, PTn), which, in turn, overlies a fine-grained, lowpermeability rock (Topopah Spring Tuff, Tpt). If matrix flow
is the only mechanism considered, as is more likely under
conditions of very low net infiltration (that component of
surface infiltration that makes it to a depth below which
evapotranspiration processes can remove it), then a capillary
barrier would exist at the interface between the fine-grained,
high water-retention matrix of the Tpc and the coarsegrained PTn. The PTn has a low relative permeability for
the range of negative water potentials likely to exist at low
infiltration rates. In addition, a permeability barrier would
exist between the base of the coarse-grained PTn and the
2. Historical Development of the Conceptual
Model of Lateral Diversion at Yucca Mountain
2.1. Conceptual Model Development
[8] The original location for a potential repository in the
unsaturated zone was first suggested by Winograd [1981].
Early project investigators had various views of the potential for lateral diversion. Scott et al. [1983] saw some
potential for capillary barrier mechanisms to be effective
at the contact of the upper welded unit, the Tiva Canyon
Tuff (Tpc), and the underlying nonwelded tuffs of the
Paintbrush Group (PTn), whereas Roseboom [1983]
claimed no lateral component would exist because of the
extensive fracture system. One year later, a conceptual
model of flow at Yucca Mountain was published by
Montazer and Wilson [1984] that included extensive discussions of capillary barrier mechanisms, and it was suggested that over 80% of the percolation (4 mm of the 4.5
mm/yr that they estimated) would be laterally diverted
above and within the PTn. This was characterized as a
large-scale process such as the scale depicted in Figure 2.
Sinnock et al. [1987], Klavetter and Peters [1986], and
Peters and Klavetter [1988] suggested the diversion would
occur at the base of the PTn due to capillary barrier
mechanisms because of the nonwelded tuff overlying even
larger pores of the fractures in the densely welded Topopah
Spring Tuff. Wilson [1996] stated that conventional wisdom
still believed that the PTn acts as an effective capillary
SBH
4-4
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
barrier to divert much of the infiltrating water around the
potential repository horizon, and, as recently as 1999, lateral
flow in and above the PTn was still being included in the
conceptual model of flow implemented in the three-dimensional site-scale flow model of Yucca Mountain [Ritcey and
Wu, 1999; Sonnenthal and Bodvarsson, 1999].
2.2. Numerical Model Development
[9] Numerical models were used to test the hypotheses
for lateral diversion using limited data and field observations in the late 1980s. Rulon et al. [1986] developed a twodimensional (2-D) model that supported Montazer and
Wilson’s [1984] concept of lateral flow in the PTn at low
fluxes, with as much as 50% of the infiltrating water being
diverted using fluxes of <1 mm/yr, but about 20% diverted
at their maximum estimate of 4.5 mm/yr, considerably less
than the nearly 90% suggested by Montazer and Wilson
[1984].
[10] Rockhold et al. [1990] ran the first 3-D model in
1990 for a small portion of the site. It produced matrix flow
in the PTn and diverted most of it laterally within the PTn.
By 1995, many of the researchers still believed that lateral
diversion in and above the PTn might reduce the volume of
water that would penetrate the Topopah Spring Tuff, even
though the estimates of potential net infiltration into the
system had been increased from the earlier estimates of 1– 5
mm/yr to an average of 5 mm/yr with a range of 0 – 80 mm/
yr distributed spatially over the site [Flint et al., 2001a,
2001b]. Although no significant or conclusive field evidence was available to support the concept of lateral
diversion, several modeling exercises were done to test
whether the available measured properties and geometry
of the PTn could support or induce diversion [Ho, 1995;
Altman et al., 1996; Moyer et al., 1996]. Typically, lateral
diversion was attained, although minimally at high fluxes
(>10 mm/yr). At lower fluxes (>0 – 5 mm/yr), the diverted
water may have been an artifact of simplified geometry,
unrealistic properties, the idealization of stratigraphic contacts as linear features, or a misrepresentation of the gradational nature of the transition of the Tiva Canyon Tuff into
the nonwelded PTn. In 1999, water was diverted laterally
above the PTn in the three-dimensional site-scale flow
model of Yucca Mountain, developed by Lawrence Berkeley National Laboratory, a distance of 250 m under
current climate conditions, with a net-infiltration rate of
0 – 38 mm/yr [Ritcey and Wu, 1999], although they
acknowledged that no field evidence of lateral diversion
has been observed and suggested that faulting or interface
heterogeneities limit lateral flow. Until only recently, there
were no accurate and high-resolution matric-potential measurements of the subsurface rocks, particularly in the PTn, to
test the model results. These measurements are now available for boreholes penetrating the PTn [Flint et al., 2002b]
and for locations along the extent of the Cross Drift [Flint
and Flint, 2000].
2.3. Lithostratigraphic Description
[11] Early interpretations of lateral diversion were on the
basis of several boreholes that provided geologic/lithologic
descriptions and scanty measurements of hydrologic properties. Once more detailed information was available, such
as detailed physical and hydraulic properties from core
samples and geochemical data, more insightful models
were developed to test the hypotheses of lateral diversion.
Figure 3 schematically illustrates the generalized lithostratigraphy of the nonwelded PTn flanked above and below
by the fractured, densely welded Tiva Canyon and Topopah
Spring Tuffs. An example of the corresponding porosity
and relative saturation of core samples from several boreholes is shown in Figure 4.
[12] The Paintbrush nonwelded hydrogeologic unit (PTn)
of Montazer and Wilson [1984] is described as consisting of
the nonwelded and partially welded base of the Tiva
Canyon Tuff, the Yucca Mountain and Pah Canyon Tuffs,
the upper nonwelded and partially welded upper part of the
Topopah Spring Tuff, and the associated bedded tuffs. It
consists of thin, nonwelded ash-flow sheets and bedded
tuffs that thin to the southeast from a maximum thickness of
200 m to a minimum of 30 m in the vicinity of the
potential repository.
[13] These rocks have been separated into several hydrogeologic units on the basis of lithostratigraphic description
[Moyer et al., 1996] and similarity in hydrologic properties
[Flint, 1998, 2002]. The rocks near the base of the Tiva
Canyon Tuff are characterized by a transition in porosity
and mineral alteration and are divided into the CMW and
CNW hydrogeologic units. The CMW hydrogeologic unit
consists of moderately welded rocks near the base of the
Tiva Canyon Tuff, with varying degrees of vapor phase
corrosion that range from 15% porosity at the top of the
unit to >28% porosity at the bottom of the unit. The CNW
consists of nonwelded to partially welded rocks at the base
of the Tiva Canyon Tuff that range from >28% porosity to
>50% porosity. The CMW and the top of the CNW in some
boreholes consist of low permeability altered minerals with
small pores and high water retention.
[14] The lithostratigraphic units of the PTn commonly are
thin but distinct enough in properties to delineate as separate
hydrogeologic units. A numerical modeling exercise was
done by Moyer et al. [1996] to assess the hydrologic impact
of these individual units and whether the mean properties of
the units were different enough to maintain the individual
layers as separate units. It was determined that abrupt and
linear contacts, along with the contrasts in properties, were
instrumental in creating lateral diversion along the sloping
contacts. As an exercise, this indicated that the contrasts in
the mean properties for each unit were different enough in
most cases to maintain separate units. Moyer et al. [1996]
suggested the possibility that the exercise simplified the
linear character of the contacts. Field observations of outcrops and analyses of core data, such as in Figure 3, and
sample properties as given by Istok et al. [1994], provide
indications that these contacts are often only locally linear.
Therefore they are less likely to divert water laterally
because of heterogeneities that cause local increases in
saturation that result in penetration across the boundary.
[15] The Yucca Mountain Tuff (TPY) has properties very
similar to the BT4 and BT3 (Figure 3) and is absent in
boreholes to the south of Drill Hole Wash (Figure 4, see
boreholes N54 and N55), but has lower porosity and becomes
moderately welded to the north (Figure 4, see borehole SD9).
The nonwelded to partially welded Pah Canyon Tuff is
hydrogeologic unit TPP. The bedded tuff BT2 and the nonwelded top of the Topopah Spring Tuff were not different
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
SBH
4-5
Figure 4. Porosity (solid circles) and saturation (open circles) for several boreholes penetrating through
the nonwelded rocks of the Paintbrush Group (PTn).
enough in properties to maintain as separate units [Moyer et
al., 1996; Flint, 1998] and are represented by hydrogeologic
unit BT2 (Figure 3). The nonwelded top of the Topopah
Spring Tuff, encompassed by BT2, is a thin unit that transitions sharply from nonwelded to densely welded rock such
that the borehole sample spacing of 0.8 m was not enough to
adequately characterize the transition. Underlying this unit is
a very thin, densely welded and fractured upper vitrophyre,
hydrogeologic unit TC (Figure 3), which is typically <0.5 m
thick, but varies from 0 to 2 m thick across Yucca Mountain.
The TC typically has porosity <5% and has microfractures
that have high permeability when saturated and very low
permeability when unsaturated. This unit also has varying
degrees of larger fractures that initiate just at the upper contact.
2.4. Evaluation of the Potential for Lateral Diversion
From Field Data and Detailed Properties
[16] Borehole data of physical and hydraulic properties
and water content collected from 1990 to 1995 provided
support for the concept of a barrier (capillary or permeability), at least in some locations (Figure 4). These profiles
of porosity and saturation suggest the possibility of barrier
mechanisms operating because of the high saturations at the
contacts between the CMW and the CNW, or the CNW and
BT4, and the abrupt decline in saturation of the underlying
rocks. The identification of high amounts of smectites in
these locations further supported the concept (D. Bish, Los
Alamos National Laboratory, written communication,
1992). The scenario suggests an initial depositional condition of a vapor phase corroded, moderately welded tuff
with small pores with a gradational transition extending
downward into a nonwelded tuff with larger pores that
overlaid, with an abrupt contact, a bedded tuff with large
pores. Given an effective barrier mechanism in this stratigraphic location, the development of long-term saturated
conditions could alter the vitric rocks to montmorillonite,
which is currently present. The presence of these altered
minerals in the overlying rocks, however, greatly reduces
SBH
4-6
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
their permeability [Flint, 1998] so that they would drain
only very slowly laterally down dip. This combined with a
much lower air-entry potential and therefore thicker capillary fringe moves the barrier upward into the fractured
welded rocks, through which the water could then drain
laterally down dip. This is supported by the modeling
exercise of Moyer et al. [1994] in which lateral diversion
occurred primarily in high-conductivity fractures of the Tiva
Canyon Tuff overlying the CMW. The mechanism, in this
case, could be either a capillary barrier or a permeability
barrier, depending on the flux mechanism (whether the
percolation flux is dominated by flow through fractures or
matrix). The locations of boreholes in which the altered
minerals are the most developed are those in topographic
positions most likely to have historically high surface
fluxes, such as washes during the high-precipitation Holocene-age climate (Figure 4, UZ7a or N31) or where surface
faulting exists (Figure 4, N55). As a result, this feature is
not well developed in all boreholes.
[17] The collection of detailed rock properties provided
information for detailed numerical modeling intended to test
the effectiveness of the PTn to produce lateral diversion.
The construction of numerical models typically relies on
mean values of properties for a given lithostratigraphic unit,
resulting in artificially abrupt contacts because of the
manner in which gradational contacts between rock types
are discretized, such as at the transition from welded to
nonwelded rocks at the top of the PTn and the moderately
welded to fractured rocks at the bottom of the PTn. In one
case, Kwicklis et al. [1994] used porosity to estimate
hydraulic properties, producing high-resolution changes in
vertical property fields in two 1-D models of boreholes. The
boreholes UZ4 and UZ5 (Figure 1) were located in a
channel and an adjacent side slope. These models resulted
in interpretations of lateral flow in the PTn from the channel
borehole to the side slope borehole, but with a hesitancy to
conclude anything more significant than subsurface spreading from a location with high surface fluxes such as this
wash, which had significant runoff in recent years, to
surrounding drier rocks.
[18] Larger-scale models consistently resulted in diversion [Moyer et al., 1994; Ho, 1995; Altman et al., 1995],
particularly at the lower fluxes provided by the range of
estimated net infiltration for the site. The use of properties
that were stochastically distributed within hydrogeologic
unit boundaries among grid cells on a fine scale, 2 cm 4
cm, appeared to reduce the large-scale diversion by 15%
[Ho and Webb, 1998]. This is probably the most realistic
representation of the geometry and properties of the site.
[19] Properties and models have supported active barrier
mechanisms and have provided support for lateral diversion.
However, isotopic data, indicating young water, such as
bomb-pulse chlorine-36 and tritium, in borehole samples
from the PTn [Civilian Radioactive Waste Management
System (CRWMS), 2000] indicate that not all water was
laterally diverted above the nonwelded units and, indeed,
that water did penetrate the potential barrier at the upper
contact. These isotopes were later found extensively in the
Topopah Spring Tuff in the Exploratory Studies Facility
[Fabryka-Martin et al., 1997], further confirming the notion
of fast paths through the PTn. All measured bomb-pulse
values were associated with faults that broke the continuity
of the PTn. The extent of paths into and through the PTn is
not completely known, however, and still does not preclude
the notion of diversion because of the PTn.
3. Potential for Diversion Due to Natural Barriers
3.1. Analytical Evaluation of a Potential Capillary
Barrier
[20] Analytical calculations can be made to evaluate the
maximum potential for lateral diversion due to capillary
barriers, assuming idealistic physical conditions. These
calculations do not consider unsaturated flow phenomena
such as fingering, or the spatial distribution of heterogeneities, that are likely to reduce diversion. This tool is therefore useful to assess when lateral diversion is unlikely to
occur, given a specific set of hydraulic properties, and may
help to ascertain the mechanisms contributing to apparent
flow conditions in the unsaturated zone.
[21] Ross [1990] published analytical calculations of
capillary barrier mechanisms leading to lateral diversion
of water at sloping interfaces. The schematic represented in
Figure 5 depicts the processes and variables used in his
analysis. The analysis assumed rather idealized conditions
to allow for the computation of lateral diversion of water. To
describe media conductivity, Ross employed Gardner’s
[1958] equation for conductivity:
K ¼ Ks eay
ð1Þ
where K is unsaturated hydraulic conductivity, Ks is
saturated hydraulic conductivity, a is a pore-size distribution index, and y is water potential. The total horizontal
flow, or maximum lateral diversion capacity (Qmax), is a
function of Ks, the angle of incline (j), the applied vertical
flux rate (q), and a for each of the two media on either side
of the interface [Ross, 1990] with the lower layer indicated
by the asterisk:
Qmax
Ks tan j
¼
a
"
q
Ks*
a #
a
q
*
:
Ks
ð2Þ
[22] Ross [1990] showed that to create a capillary barrier
with significant diversion capacity, fine-textured media
must overlie coarse-textured media with a linear, sloping
contact, and a low flux rate. The barrier effect is overcome
when the saturation is high enough in the overlying layer for
the capillary pressure to be more than the underlying airentry pressure. The barrier exists because of the difference
in effective permeability between the fine layer and the
coarse layer under unsaturated conditions. The length of
diversion L is a function of the upper layer only in the work
by Ross [1990]:
L<
Ks tan j
:
aq
ð3Þ
[23] Steenhuis et al. [1991] extended this equation by
making it a function of the air-entry pressure of the upper
layer and the water-entry pressure (about equal to half the
air-entry pressure) of the lower layer, which resulted in
about twice the length of diversion. These calculations have
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
Figure 5. Schematic of analytical calculations of lateral
diversion (modified from Selker et al. [1999, Figure 3.9])
where Q is diversion capacity (cm2/d), Qmax is maximum
diversion capacity (cm2/d), L is diversion length (cm), q is
applied flux (cm/d), f is angle of incline, a is pore size
indicator (1/cm) using Gardner [1958], and Ks is saturated
hydraulic conductivity (cm/d).
been shown to provide reasonable results in controlled
laboratory experiments [Walter et al., 2000].
[24] Ross’s [1990] equations assume an infinite thickness
of the upper and lower layers isolating the interface from
water potential gradient influences. This assumption is
violated in gradational transitions depending on how the
media geometry is represented. At Yucca Mountain the
small pores of the altered, moderately welded Tiva Canyon
Tuff have a very large air-entry value (calculated as 1/a),
creating a saturated capillary fringe much thicker than the
extent of the unit thickness, which averages about 2– 9 m.
Ross [1990] provides a simplified approximation for this
case, which neglects the contrast with the lower layer and
incorporates no influence of applied flux q where b is unit
thickness:x
Qmax ¼ Ks b sin j:
ð4Þ
[25] However, as the pore size of the altered, moderately
welded rocks is gradational with distance, there is no
distinct upper bound to the layer. In essence, as the pores
become smaller upward from the sloping interface, they
provide a continuous water column and capillary fringe,
until the height at which the transition is within the
moderately welded to welded rocks and fractures are able
to drain the accumulating water in the capillary fringe down
dip. This conceptualization diffuses the clarity of the appropriate geometry or the most appropriate equation to use for
calculations of Qmax, but supports the contention that either
analytical calculation will serve as a maximum potential
diversion.
[26] Several analyses were done to evaluate the maximum
potential for lateral diversion by using equations (2) and (4)
to calculate Qmax. Given the mean measured properties at
this site, a 7-degree slope, the mean flux and range of
SBH
4-7
fluxes, and the scale at which the geometry of features exist,
several potential conditions are evaluated to assess the
likelihood of any diversion extending beyond the dimensions of the potential repository, about a maximum of 1000
m down dip. In the following sections, four issues will be
considered for analysis.
1. The first issue involves the upper transitional zone.
Could there be lateral diversion within the transition of the
CMW through the CNW? (1) What is the relation between
number of layers used in the representation of the
gradational system (corresponding to degree of contrast
between adjacent layers) and the amount of diversion?, (2)
how high must the saturated hydraulic conductivity be in
the CMW to get appreciable diversion (conductivity is
correlated to the development of alteration in the CMW)?,
and (3) how low does the applied flux rate have to be for the
given set of properties to predict significant lateral
diversion?
2. The second issue involves an historical scenario.
Could the conditions suggested in the historical scenario
have sustained a capillary barrier to potentially result in the
altered CMW, and how do the results compare with
calculations of the existing transition?
3. The third issue involves the hydrogeologic units of the
PTn. Are there any contrasts in properties of the units within
the PTn that might produce lateral diversion?
4. The final issue involves the lower transitional zone. Is
there a capillary barrier due to the position of the lower
transition of the moderately welded Topopah Spring Tuff
over an assumed range of fracture properties in TC?
3.1.1. Upper Transitional Zone
[27] The first analysis to evaluate whether the CMW/
CNW transition is effective in promoting lateral diversion is
set up to test whether unrealistic contrasts in properties
imposed by typical model development will result in
inadvertent diversion. Average conditions and properties
are used with a 9-m-thick unit that transitions from a
porosity of 0.14 to 0.38. Values of Ks and a are correlated
with porosity [Flint, 2002] from measured data to provide
realistic contrasts between the layers. Hydraulic conductivities for six of the different porosities are shown in Figure
6a. Computations of maximum expected lateral diversion
Qmax are made using equations (2) and (4) on the basis of an
average flux of 5 mm/yr (0.00014 cm/d), encountering
between two and 12 layers (Table 1). Qmax1 uses equation
(2), which allows us to examine the relative maximum each
layer could divert. Qmax2 is based on equation (4), assuming a limitation imposed by the layer thickness violates the
assumptions for equation (2).
[28] An assessment of the relation of number of layers to
calculated Qmax shows that even though the two-layer
system has the largest contrast in layer properties, the lower
hydraulic conductivity of the layer with a porosity of 0.14
limits the lateral diversion (Table 1 and Figure 6b). The
three-layer system has a somewhat less, but still large,
contrast between two of the layers, but the higher hydraulic
conductivity of the layer with a porosity of 0.26 allows
much more lateral diversion. Reducing the contrast in a and
Ks by including additional layers reduces the diversion from
this maximum for Qmax1.
[29] To evaluate the sensitivity of the parameters and the
relation of Ks and q to Qmax, fluxes from 1 to 20 mm/yr
SBH
4-8
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
Figure 6. Data from calculations of lateral diversion potential Qmax (a) hydraulic properties calculated
for six porosities, (b) Qmax calculated for systems with different numbers of layers, (c) Qmax versus the
ratio of q to Ks for five applied fluxes, and (d) Qmax and the ratio of a for the upper layer (a) to the lower
layer (a*) for a range of Ks.
were used to calculate Qmax. Figure 6c indicates that an
optimum ratio of q to Ks of approximately 0.01 to 0.0001
results in the highest Qmax (>2 cm2/d) for the given set of
properties. While the largest contrast possible in a for the
two layers will produce the highest Qmax (Figure 6d), there
is an optimum Ks for the given set of properties in Table 1.
For example, for a flux of 0.5 mm/yr (0.00014 cm/d), a Ks
of near 0.1 cm/d results in the highest Qmax.
[30] Using the available analytical tools, we can also
investigate diversion calculated assuming the most idealistic
geometry exists, which provides the highest possible calculated Qmax. When equation (4) is used for Qmax2, which
assumes the capillary fringe exceeds the layer thickness, the
calculation is not limited by the upper boundary flux and the
Qmax is primarily a function of Ks, which is relatively high
for more layers, nor is there any breakthrough into a lower
layer. The calculation of Qmax2 assumes that the layer
thickness is much less than (a cos j)1, which is violated
by most of the layers that have a thickness between 3% and
90% of the calculation of (a cos j)1. There can be no
direct comparison between Qmax1 and Qmax2, but they
provide bounds for high and low diversion, given the
possible interpretations of the physical system and the
applicability of the assumptions.
[31] To produce a more realistic set of Qmax values, Qmax1
(multiplying by layer thickness/air entry) was scaled to
reflect the actual layer thickness that would provide a
saturated capillary fringe flowing laterally down dip. These
4-9
SBH
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
Table 1. Calculations of Lateral Diversion Potential and Distance for a Sloping Interfacea
Qmax1
2 layers
3 layers
4 layers
5 layers
6 layers
8 layers
10 layers
12 layers
Porosity
Layer Thickness, cm
0.14
0.38
0.14
0.26
0.38
0.14
0.22
0.30
0.38
0.14
0.20
0.26
0.32
0.38
0.14
0.19
0.24
0.28
0.33
0.38
0.14
0.17
0.21
0.24
0.28
0.31
0.35
0.38
0.14
0.17
0.19
0.22
0.25
0.27
0.30
0.33
0.35
0.38
0.14
0.16
0.18
0.21
0.23
0.25
0.27
0.29
0.31
0.34
0.36
0.38
450
450
300
300
300
225
225
225
225
180
180
180
180
180
150
150
150
150
150
150
113
113
113
113
113
113
113
113
90
90
90
90
90
90
90
90
90
90
75
75
75
75
75
75
75
75
75
75
75
75
1
a, cm
0.0004
0.0124
0.0004
0.0018
0.0124
0.0004
0.0009
0.0035
0.0124
0.0004
0.0007
0.0018
0.0049
0.0124
0.0004
0.0006
0.0012
0.0027
0.0059
0.0124
0.0004
0.0005
0.0008
0.0013
0.0024
0.0042
0.0074
0.0124
0.0004
0.0005
0.0006
0.0009
0.0014
0.0022
0.0035
0.0055
0.0083
0.0124
0.0004
0.0005
0.0006
0.0008
0.0010
0.0015
0.0021
0.0031
0.0045
0.0064
0.0090
0.0124
Qmax2
Air Entry, cm
Ks, cm/d
Per Layer
Total
Per Layer
Total
L, m
2278
80
2278
567
80
2278
1084
286
80
2278
1423
567
204
80
2278
1634
846
375
168
80
2278
1861
1274
756
422
236
135
80
2278
1975
1541
1084
710
451
286
183
120
80
2278
2042
1708
1328
971
682
470
323
224
157
111
80
0.001
2.069
0.001
0.128
2.069
0.001
0.038
0.366
2.069
0.001
0.019
0.128
0.587
2.069
0.001
0.012
0.063
0.245
0.769
2.069
0.001
0.007
0.025
0.078
0.205
0.481
1.034
2.069
0.001
0.005
0.015
0.038
0.087
0.185
0.366
0.683
1.214
2.069
0.001
0.004
0.010
0.023
0.048
0.094
0.173
0.306
0.518
0.846
1.341
2.069
0.2
0.2
0.1
0.1
17
0.0
2.3
2.3
0.1
4.7
4.7
2
166
0.0
0.6
0.9
1.5
0.0
1.0
10.0
11.1
0
44
62
0.0
0.2
0.4
0.3
0.9
0.0
0.4
2.8
12.9
16.1
0
14
31
24
0.0
0.1
0.2
0.2
0.2
0.7
0.0
0.2
1.2
4.5
14.1
19.9
0
5
16
17
11
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.4
0.0
0.1
0.3
1.1
2.8
6.6
14.2
25.1
0
1
5
8
8
6
4
0.0
0.0
0.0
0.1
0.1
0.1
0.0
0.0
0.0
0.3
0.0
0.1
0.2
0.4
1.0
2.0
4.0
7.5
13.3
28.4
0
0
2
4
5
4
4
3
2
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.0
0.0
0.1
0.2
0.4
0.9
1.6
2.8
4.7
30.7
0
0
1
2
3
3
3
3
2
a
A slope of 7 degrees is used with an applied flux of 0.5 mm/yr. Qmax1 is diversion potential using equation (2), Qmax2 uses equation (4) for a layer
with an upper boundary, and L is calculated as Qmax1 divided by the applied flux.
resulting values are 20– 50% lower than Qmax1 (Figure 6b).
Regardless of the assumptions for the calculations of Qmax,
they represent the highest realistic potential for lateral diversion in this gradational unit transitioning from the welded
rocks to the nonwelded rocks (Table 1), and yet they show
virtually no lateral diversion when the system is discretized to
include more than about five layers in 9 m of thickness.
[32] In equation (3), Ross [1990] assumes that the pore
size distribution index a provides an estimate for the
capillary fringe height, which is necessary for calculating
diversion length L. The assumption of an infinite upper
boundary distance is violated for our purposes; estimates of
L can be simplified by dividing Qmax1 (in cm2/d) by q (in
cm/d) to attain L (in cm). These results (Table 1) indicate
that water is not diverted laterally distances that approach
the dimensions of the potential repository. The distances are
insignificant when the system is discretized to include more
than six layers, but even the greatest distance of 166 m in
the three-layer system is relatively inconsequential considering the scale of the repository.
3.1.2. Historical Scenario
[33] To evaluate a historical scenario prior to the extensive mineral alteration of the base of the Tiva Canyon Tuff,
a surrogate set of properties is considered. The top of the
Tiva Canyon Tuff has a somewhat parallel development to
that of the bottom, with similar welding due to cooling rate,
and vapor phase corrosion influencing the porosity. While
the upper part is crystal rich, rather than crystal poor, there
SBH
4 - 10
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
Table 2. Mean Values and Standard Deviations for Measured Core Properties, Compiled From All Boreholes for Each Hydrogeologic
Unita
Porosity (v/v)
Saturated Hydraulic
Conductivity, m/s
Saturation
Moisture-Retention
Curve-Fit Parametersb
Hydrogeologic
unit
Mean
Standard
Deviation
Mean
Standard
Deviation
N
Geometric
Mean
Standard
Deviation
N
a,
MPa1
n
m
N
Fracture
Density (F/m)
CW
CMW
CNW
BT4
TPY
BT3
TPP
BT2
TC
0.08
0.20
0.39
0.44
0.27
0.41
0.50
0.49
0.05
0.03
0.05
0.07
0.12
0.09
0.08
0.04
0.10
0.04
0.80
0.91
0.69
0.51
0.64
0.53
0.36
0.39
0.63
0.14
0.13
0.24
0.20
0.23
0.16
0.13
0.15
0.17
616
96
105
34
48
87
166
177
72
5.7E-12
1.8E-10
1.2E-08
5.8E-07
1.6E-07
5.4E-07
9.3E-07
2.2E-06
1.6E-09
3.4E-11
1.3E-07
1.1E-05
2.1E-03
7.8E-06
1.4E-06
4.4E-07
9.4E-06
2.6E-07
16
5
10
4
3
17
11
21
5
2.8
0.3
96.2
53.3
162.6
93.2
47.2
51.4
1.5
1.20
1.35
1.19
1.21
1.25
1.16
1.36
1.23
1.27
0.17
0.26
0.16
0.18
0.20
0.14
0.26
0.19
0.21
9
3
6
7
3
5
2
7
3
5.0
1.0
0.5
0.5
1.0
0.5
1.0
0.5
25.0
a
N, number of samples; MPa, megapascals; F/m, fractures per meter; v/v, dimensionless volume.
Moisture-retention curve-fit parameters determined using van Genuchten [1980].
b
are similarities in hydrologic properties [Flint, 2002]. If the
properties of the crystal-rich moderately welded rock (a =
0.0012 cm1, Ks = 0.138 cm/d) are used to overly the CNW,
the Qmax is 46 cm2/d compared with 0.24 cm2/d in the
existing rocks. This suggests that capillary conditions are
viable to creating saturated conditions conducive to mineral
alteration in vitric rocks. Once the minerals are altered, the
rocks then maintain high saturation due to moisture-retention characteristics.
[34] This evaluation provides credence to the early conceptual models that were based on minimal rock property
data and general lithostratigraphic descriptions. It was not
until detailed information, including measurements of the
properties and mineral alteration of the gradational transition, was available that this conceptual model would lose its
viability.
3.1.3. Hydrogeologic Units of the PTn
[35] Early conceptual models also suggested that lateral
diversion occurred within the layers of the PTn because of
the apparent linearity of the contacts over long distances. It
was assumed that the properties were similar spatially
within the units as the deterministic depositional processes
governed their features [Rautman and Flint, 1992; Istok et
al., 1994; Flint et al., 1996]. Mean properties for the
nonwelded ash-fall and bedded tuff layers in the PTn are
relatively similar, but with large variability over the site
(Table 2). This spatial variability was confirmed by Istok
et al. [1994], Moyer et al. [1996], and Flint [1998, Table
8]. Generally the units within individual boreholes because
of the nature of the depositional process do not have
abrupt changes in properties (Figure 4). However, calculations could be made using the specific properties for any
given borehole and get widely variable results. Calculated
lateral diversion Qmax from the mean values results in
about 4 cm2/d for the contrast between BT4 and TPY and
no diversion for the contrasts between the other adjacent
units, even within the ranges of their variance. (The TPY
only exists north of the potential repository location and
thus is not very relevant to the consideration of lateral
diversion above the repository.) The key here is that the
lateral variability between boreholes is high enough that
lateral diversion is very unlikely to extend any distance
within the PTn units. This is particularly evident on the
basis of the fracturing in the units (Table 2) and faults that
result in small offsets that interrupt the linear continuity
[Heiberger, 1996]. The fracture density for the PTn hydrogeologic units is one fracture every 1– 2 m, and while these
fractures are not likely to result in offsets creating contrasts
in properties, they are likely to conduct water faster than the
matrix. The consequences of the faults that break the PTn
are evident on the basis of numerous measurements of
bomb-pulse chlorine-36 in the Exploratory Studies Facility
(ESF) [Fabryka-Martin et al., 1997]. These measurements
indicate that not only are there breaks in the PTn, but
continuous fracture paths extend from the surface through
the PTn to the ESF in many locations. The presence of
bomb-pulse chlorine-36 indicates the presence of water that
is less than about 50 years in age, inferring that a fast path
for the water to percolate through the PTn must exist. The
frequent spacing of these bomb-pulse chlorine-36 measurements in the ESF supports the conceptual model of the PTn,
which notes that it is broken by faults and does not have the
lateral continuity necessary for large-scale diversion.
[36] One approach to analyzing the variability in the units
and therefore the varying contrasts between the layers, is to
stochastically distribute the properties in each unit using the
mean and variation for each unit and to evaluate the range in
contrasts at the boundaries between units. Modeling exercises have investigated diversion using this approach with
resulting declines in diversion in comparison to a simple
layered approach to distributing the properties [Altman et
al., 1995; Ho and Webb, 1998].
3.1.4. Lower Transitional Zone
[37] The base of the PTn is the final avenue for lateral
diversion above the potential repository horizon. This
transition ranges from a porosity of about 0.4– 0.05 or less
over a distance of about 1 – 2 m. The porosity transition
within the BT2 is smooth; however, the contact with the
vitrophyre, TC, which has many very small to relatively
large fractures, is very abrupt. The analysis, in this case, is
not as in the transitional example, but considers two-layer
scenarios. Properties for fractures of specific apertures were
modeled by Kwicklis and Healy [1993] and are used in this
analysis. To derive a values for the calculations, a correlation of a using Gardner [1958] and a from van Genuchten
[1980] were made on the basis of 18 samples from those
used to calculate the mean values in Table 2, which
represented the range of units from welded to nonwelded
SBH
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
4 - 11
Table 3. Properties for Calculation of Lateral Diversion Potential and Length of Lower Paintbrush Tuff Nonwelded Transition Into
Topopah Spring Tuff Welded Rocksa
Matrix Porosity
(v/v)
Vitric tuff
Fractures
Vitric tuff
Fractures
Vitric tuff
Fractures
Vitric tuff
Fractures
Vitric tuff
Fractures
Vitric tuff
Fractures
Fracture Aperture,b
microns
0.2
25
0.2
125
0.1
25
0.1
125
0.05
25
0.05
125
Gardner a,
cm1
Ks,
cm/d
Air Entry,
cm
Total Qmax,
cm2/d
L,
m
0.0007
0.0248
0.0007
0.1482
0.0004
0.0248
0.0004
0.1482
0.0004
0.0248
0.0004
0.1482
1.9E-02
1.8E-02
1.9E-02
2.7E-01
1.2E-04
1.8E-02
1.2E-04
2.7E-01
7.2E-07
1.8E-02
7.2E-07
2.7E-01
1423
40
1423
7
1423
40
2465
7
2499
40
2499
7
2.8
207
3.2
230
0
0
0
0
0
0
0
0
a
Ks, saturated hydraulic conductivity; Qmax, diversion potential; L, length; v/v, dimensionless volume; cm, centimeters; m, meters.
Fracture properties from Kwicklis and Healy [1993].
b
for which curve-fit parameters were estimated using both
equations. The correlation had an R2 of 0.89.
[38] Properties for nonwelded tuff of three porosities
were used to overlie fractures of two apertures (Tables 3
and 4). As the contrast between a is high for all sets of
layers, the Ks is the limiting factor in the upper layer for the
0.1 and 0.05 porosity rocks. The layers with porosities of
0.05 and 0.1 have zero diversion length and no diversion
capacity, whereas the layer with a porosity of 0.2 has a
potential diversion of greater than 200 m. The difference in
fracture aperture makes little difference in the calculation of
Qmax or L. These results show the importance of accurately
representing the upper layer properties in the calculation of
lateral diversion at this transitional boundary. They support
the contention that lateral diversion caused by a capillary
barrier in this transitional zone is insignificant when the
system geometry is realistically represented.
3.2. Evaluation of Large-Scale Lateral Diversion
[39] Large-scale water flow processes can be investigated
using measurements of rock samples collected from surfacebased boreholes or subsurface excavations, or monitoring of
moisture conditions in deep boreholes within subsurface
excavations. These measurements represent vertically and
laterally integrated subsurface conditions at various scales
[Flint et al., 2002b] and assist in the evaluation of processes
operating at the scale of the potential repository. The
analyses of these data lend evidence of generally vertical
flow through the PTn whether via fast paths or through the
matrix. These apparently dominant processes do not reduce
the amount of water percolating through the repository
horizon.
3.2.1. Subsurface Collection of In Situ Water
Potential Data
[40] To evaluate large-scale lateral diversion, Darcy’s law
was used to calculate the lateral flux downslope between
two boreholes drilled vertically penetrating the PTn and the
upper rocks of the welded Topopah Spring Tuff (Figure 7).
This method uses estimates or measurements of hydraulic
conductivity of the rock at its prevailing state of water
potential or saturation, in combination with estimates of the
hydraulic-head gradient, to directly calculate percolation
flux using Darcy’s law. Two alcoves were mined into the
wall of the north ramp of the Exploratory Studies Facility
(ESF) (Figures 1 and 7) 770 m apart. The west dipping
north ramp slopes downward through all units of the PTn.
Boreholes were drilled in the floor of each alcove and
instrumented with heat dissipation probes [Flint et al.,
2002a] located to measure the matric potential in the various
nonwelded and bedded tuffs. Hydraulic properties were
measured on samples collected from the boreholes during
drilling.
[41] Volumetric water content was determined with neutron moisture meters and laboratory measurements of core,
and the first 5 – 14 m of the two boreholes showed the
effects of evaporation from ventilation in the drift. Flux was
calculated using matric potential measurements deeper in
the boreholes away from this influence, following equilibration for over a year. Flux was calculated to be downward
in the boreholes at rates of 8 – 15 mm/yr in the nonwelded
PTn and 1 mm/yr in the underlying Topopah Spring Tuff
densely welded rocks. These flux estimates suggest that a
significant part of the flux through the PTn is either laterally
diverted or converted to fracture flow and therefore does not
influence the matric potential of the rock matrix when it
reaches the more welded unit below. Calculations of flux
along the 6.5-degree slope between the two boreholes
resulted in less than 1 mm/yr down dip flux in either the
base of the PTn or in the welded top of the Topopah Spring
Tuff, supporting the suggestion that most of the water is
converted to fracture flow once it penetrates the welded
rocks. The uncertainties in this approach include those
Table 4. Chloride Concentration (mg/L) for Core Samples From
Seven Boreholes Located on the Ridge at Yucca Mountain and
Down Dip From the Ridgea
SD-6
CNW
BT4
TPY
BT3
TPP
BT2
a
43
30
SD-9
170
138
93
64
40
SD-7
SD-12
UZ-7a
50
70
77
46
60
NRG-7a
NRG-6
148
39
54
74
Hydrogeologic unit names of the PTn are described in Figure 3.
58
62
49
SBH
4 - 12
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
Figure 7. Schematic of instrumented boreholes in alcoves 3 and 4 in the North Ramp of the Exploratory
Studies Facility (ESF). The North Ramp slopes downward to the west through the eastward sloping beds
of the nonwelded rocks of the Paintbrush Group (PTn).
encountered in the measurement of hydraulic conductivities
for samples with low saturations, and the calculated fluxes
are accurate only within an order of magnitude. Nevertheless, it provides strong indication of a lack of significant
lateral diversion down dip over this distance of 700 m. It
also provides support for the interpretation of the onedimensional model results [Kwicklis et al., 1994] that
suggest that local spreading occurs in the PTn in locations
of concentrated surface runoff or high net infiltration rather
than extensive lateral diversion.
[42] A larger-scale evaluation of subsurface distribution
of moisture can be used to assess the likelihood of lateral
diversion due to the PTn. A comparison of measured matric
potential in the Cross Drift (Figure 1) with modeled surface
net infiltration (Figure 8) was used to illustrate the apparent
lack of significant diversion. Net infiltration is the part of
water that crosses the air-surface interface and penetrates
deeper than plant roots and evaporation processes and
therefore may be assumed to equal percolation flux within
the unsaturated zone. Net infiltration is modeled using a
water-balance approach, detailed measurements of distributed precipitation, evapotranspiration, and soil and rock
water content and properties [Flint et al., 2000]. The
comparison is shown for measured matric potential in the
Cross Drift with estimated net infiltration at the ground
surface along the trace of the Cross Drift (Figure 9). The
highest infiltration rate over the Cross Drift is in the PTn
where it is exposed at the ground surface on the west side of
Yucca Mountain (Figure 2), and is notable in Figure 9
between stations 2200 and 2400. Further west of the
exposure of the PTn, the infiltration rate ranges from 1
to 10 mm/yr and enters the mountain directly in the Topopah Spring Tuff. This yields a rock matric potential in the
Cross Drift at station 2450 to 2700 of slightly higher than
0.1 MPa (Figure 9). In this case, there is no overlying PTn
and no potential for lateral diversion. The modeled infiltration flux exceeds the saturated hydraulic conductivity of the
tuff matrix; therefore fracture flow is dominant and the
small fracture/matrix interaction (estimated to be 0.01% of
the total contact area by Sonnenthal and Bodvarsson
[1999]), and matrix hysteresis does not allow the matrix
to come into potential equilibrium with the fractures. This
analysis does assume the wetter rock matrix is associated
with greater flux through the fracture network. There are
many subtleties in Figure 9 that, when evaluated further,
may give additional insight into the large-scale mechanisms
responsible for subsurface flow. One such subtlety is a shift
of lower infiltration (<1 mm/yr) at station 2150 m, which
may correspond with the drier rock (>0.1 MPa) at 1700 m.
If the assumption of wetter rock matrix corresponding to
greater subsurface flux is correct, this may denote lateral
diversion of water that infiltrates above station 1700 down
dip, and no diversion of the water that infiltrates to the west
of station 2200; thus no flux enters the Topopah Spring Tuff
above station 1700. The higher infiltration of >10 mm/yr
near station 1700 may correspond to the wetter rock at
station 1300. This shift may be due to a 500– 600 m of
lateral diversion. The overall trend, however, particularly
from 0 to 1000 m, shows drier rock under lower infiltration
and wetter rock under higher infiltration, suggesting little
lateral diversion. There is little or no infiltration west of the
end of the Cross Drift because of thick alluvial deposits, so
the only water available is water from directly above,
further supporting the argument against lateral diversion
caused by the PTn.
3.2.2. Estimates of Percolation Flux in Topopah
Spring Tuff
[43] There are additional lines of evidence supporting
vertical flow through the PTn. In an evaluation of largescale flux processes, if there is a relative similarity between
estimates of net infiltration at the ground surface and
percolation flux in or below the PTn, it is assumed that
the PTn is not diverting a significant part of the percolation
water that would ordinarily flow downward through the
repository horizon.
[44] In an analysis of methods used to estimate recharge
or percolation flux at Yucca Mountain, Flint et al. [2002b]
describe and show a temperature profile analysis for two
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
SBH
4 - 13
Figure 8. Average annual net infiltration spatially distributed at the surface of Yucca Mountain, Nevada,
with potential repository boundary, underground tunnels, and boreholes used in large-scale analyses
indicated.
boreholes. The temperature profile for a borehole is simulated numerically using measured thermal properties, average annual ground-surface temperatures estimated for each
borehole, and measured temperature of the groundwater in
the upper part of the saturated zone. The percolation flux for
the model is varied to optimize the fit between the measured
and simulated temperature profiles. These analyses, conducted on that part of the borehole below the PTn, indicate a
Figure 9. Comparison of estimated net infiltration at ground surface with water potential measured
subsurface in the Cross Drift.
SBH
4 - 14
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
high flux (10 mm/yr) directly under the crest of Yucca
Mountain for one borehole and a lower flux of less than 1
mm/yr for the borehole to the east of the crest in a wash.
These estimates reasonably match the values estimated from
the modeled net infiltration for those locations and are
shown along with results from eight additional borehole
analyses in Figure 10. The estimates of flux generated from
modeling temperature profiles somewhat underestimate the
flux predicted from the net-infiltration model but are of the
same order of magnitude, supporting the similarity in surface and subsurface fluxes.
[45] Estimates of average percolation rate through the
unsaturated zone also have been made using pore-water
chloride (Cl) concentrations and the chloride-mass-balance
method. This method assumes that the flux of Cl deposited
at the surface equals the flux of Cl carried through the
unsaturated zone by infiltrating water [Fabryka-Martin et
al., 1994]. Percolation rates I are estimated from measured
Cl concentration using the relationship I = (PCo)/Cs, where
P is average annual precipitation; Co is the effective average
Cl concentration in precipitation, including the contribution
from dry fallout; and Cs is the measured Cl concentration in
subsurface water. The highest percolation rates (4 – 10 mm/
yr) are estimated from data collected in the ESF and the
Cross Drift that occur beneath areas with negligible soil
cover, such as the ridge tops and side slopes, which is
confirmed by surface estimates of net-infiltration flux rate
(Figure 11). The lowest percolation rates (0– 1 mm/yr) are
estimated from samples collected from boreholes in washes
with thick alluvium. The flux estimates are from net
infiltration simulated on the basis of 30-m grid cells. To
capture the increase in scale necessary to compare the
surface infiltration with percolation flux in the PTn or
Topopah Spring Tuff, the bounds in Figure 11 represent
the range of net-infiltration flux over an area surrounding
the borehole location of 200 m in diameter. The samples
were collected from the PTn in all of the boreholes and in
the North and South Ramps of the ESF, and from the
Topopah Spring Tuff in the Main Drift and Cross Drift of
the ESF. Estimates of percolation within the PTn that are of
the same order of magnitude as those estimated at the
ground surface specifically support a lack of diversion
because of the upper transition from welded to nonwelded
rocks. The calculations of percolation flux using samples
from the Topopah Spring Tuff of the repository horizon are
also of the same magnitude as that simulated at the surface.
The majority of the measurements and calculations of flux
in the subsurface using the temperature simulations and
CMB calculations support the surface estimates of net
infiltration and generally negate a substantial influence of
the PTn in diverting water from the repository horizon.
4. Summary and Conclusions
[46] On the basis of the field data, interpretations, and
calculations above, it seems clear that the early conceptual
models of lateral diversion at Yucca Mountain did not take
into consideration the scale at which the mechanisms
responsible for lateral diversion operate in a natural system.
Nor did they consider the range in surface fluxes possible
from surface heterogeneities. Neither data nor field observations corroborate the existence of lateral diversion caused
by a barrier effect at the bottom of the Tiva Canyon Tuff;
Figure 10. Percolation flux simulated from net infiltration
model and estimated to simulate measured temperature
profiles in several boreholes.
however, observations on the basis of boreholes samples of
nearly saturated conditions abruptly declining to <50%
saturation supports the possibility of localized barrier
effects. The current conceptual model of flow through the
PTn describes flow as vertical and slow through the PTn
matrix, but there possibly may be local-scale lateral diversion at linear contacts or above low-permeability layers.
4.1. Support of Conceptual Model From Calculations
[47] Calculations of the potential for lateral diversion
potential do not support significant volumes of diverted
water even on the basis of an idealistic representation
using mean layer properties. The volume of diverted water
is insignificant, and idealized calculation of the maximum
lateral distance that water could be diverted is <200 m
when system discretization is coarse and <10 m when
discretization is fine. None of these estimates approaches
the down-dip dimensions of the potential repository, which
is 1000 m. The volumes of diverted water are likely to
be even lower when the spatial variability of the properties
is considered, as well as the presence of fractures and
faults within the hydrogeologic units of the PTn. Additional calculations using Darcy’s law made from point data
of subsurface water potentials support downward flux in
the PTn but suggest that lateral flow over a 700-m
distance is less than 1 mm/yr in an area where the overall
flux is >5 mm/yr.
[48] These analyses suggest that lateral diversion caused
by barriers is a small-scale process and under natural
conditions is very unlikely to occur on a large scale,
especially in volcanic environments where local heterogeneities, faulting, and fractures predominate. Even apparently
uniform ash-flow and ash-fall units have a spatial distribution of hydraulic properties that result in a range of contrasts
between adjacent units that limit large-scale capillary barrier
mechanisms and permeability barriers to local effects.
Because the estimates of diversion are very sensitive to
the properties used in calculations of diversion caused by a
capillary barrier, the details of the features and properties
are necessary to accurately represent and predict these
effects. An alternative approach is to stochastically represent the distribution of properties from the mean values and
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
SBH
4 - 15
Figure 11. Flux calculated using the chloride-mass-balance (CMB) method for 12 boreholes and
locations in the Exploratory Studies Facility (ESF) and Cross Drift. All samples are from the PTn, except
for those from the ESF Main Drift and the Cross Drift, which are Topopah Spring Tuff.
the variance, and then estimate the range of diversion
possible from numerical models.
4.2. Evidence for the Lack of Large-Scale Lateral
Diversion Caused by PTn
[49] The evaluation of the unsaturated zone at Yucca
Mountain has been done using methodologies that represent
various spatial scales. Estimates of percolation flux in the
Topopah Spring Tuff and other rocks deeper than the PTn
generally represent large-scale estimates of flux integrated
over large vertical distances. If the early conceptual models
of lateral diversion caused by the barrier mechanisms
imposed by the PTn were correct, water would be diverted
laterally above the repository horizon in the Topopah Spring
Tuff and the percolation fluxes would be small in comparison to the fluxes at the ground surface.
[50] An analysis of the distribution of matrix potential in
the Cross Drift indicates a general correlation of zones with
less negative matric potential and zones with estimates of
high net infiltration at the surface. Zones with a more
negative matric potential, indicating drier rock, generally
corresponded to lower net-infiltration rates. Comparisons of
percolation flux estimates using ambient temperatures of
deep boreholes, and chloride-mass-balance calculations
using data collected in the subsurface in or at depths greater
than the PTn, with estimates of surface net infiltration,
illustrated similar magnitudes of flux at the surface and
below the PTn. These observations support the lack of
large-scale diversion caused by the PTn.
[51] There is no one approach, analysis, or data set that
provides the necessary support to refute or fully support that
lateral diversion occurs above the repository horizon at
Yucca Mountain. The combination of analyses and data
interpretations of various scales provides evidence to conclude that lateral diversion probably occurs in localized
areas, perhaps of the order of meters. However, in addition
to the data analyses presented, several other observations
provide support for a lack of large-scale diversion. These
observations include the spatial variability of matrix properties, which results in discontinuous contrasts at unit interfaces, and the pervasive presence of fractures that obviate
the entire capillary barrier process. Diversion is necessarily
limited in distance along unit interfaces to the extent that
fractures occur in the PTn, or in the rocks above and below
the PTn. Although lateral diversion above the repository
horizon may occur as a result of permeability barriers at the
top and the base of the PTn, the analyses presented that
represent large-scale processes provide the same support for
generally downward flow. As a result, it would be imprudent to rely on significant lateral diversion to maintain a dry
subsurface or reduce the downward percolation of water
through the repository horizon.
[52] There are several obvious gaps in the detailed knowledge of information at Yucca Mountain that would be
prudent to fill were these processes to be unequivocally
assessed. Properties of the fractures at the transitions from
welded to nonwelded rocks at the top of the PTn, as well as
the transition from the PTn into the TSw, are unknown. It is
not known what the distribution of apertures is, or how
pervasive the fracture filling is for these locations. Assessments of the lateral heterogeneity of the properties of the
PTn did not provide enough details to assess the variability
in contrast of hydraulic properties between layers from an
analytical perspective and did not address the distribution of
fractures penetrating the upper part of the potential barrier at
the top of the PTn.
[53] The potential for lateral diversion in natural environments is critical to the understanding of moisture flux and
distribution that is required by many scientific investigations, and the application of concepts developed at this site
should be broad. The existing state of the knowledge for
adequate representation of water flow mechanisms in com-
SBH
4 - 16
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
plex and heterogeneous systems such as fractured rock is
nowhere near complete. Theoretical analyses should be
pursued to provide accurate and reproducible analytical
solutions representing the hydraulic processes in fractured
rock, fracture/matrix interaction, and transitions into layers
with few fractures. The studies at Yucca Mountain have
produced volumes of information, data, and observations to
apply to the development of the basic theory and numerical
representation of these processes. Although numerical models can be used to further evaluate the role of variable fluxes
and the presence of fractures, analytical tools should be
developed to address the theory of the transition from
matrix flow to fracture flow and subsequent transition from
capillary barriers to permeability barriers.
[54] Acknowledgments. We thank two reviewers, Benjamin Ross
and Edward Kwicklis, as well as two anonymous reviewers, for excellent
suggestions for improving the manuscript.
References
Altman, S. J., B. W. Arnold, R. W. Barnard, G. E. Barr, C. K. Ho, S. A.
McKenna, and R. R. Eaton, Flow calculations for Yucca Mountain
groundwater travel time (GWTT-95), Rep. SAND96-0819, Sandia Natl.
Lab., Albuquerque, N. M., 1996.
Buesch, D. C., R. W. Spengler, C. T. Moyer, and J. K. Geslin, Proposed
stratigraphic nomenclature and macroscopic identification of lithostratigraphic units of the Paintbrush Group exposed at Yucca Mountain, Nevada, U.S. Geol. Surv. Open File Rep., 94-469, 47 pp., 1996.
Civilian Radioactive Waste Management System (CRWMS), Managing and
Operations contractor, Analysis of geochemical data for the unsaturated
zone, Rep. ANL-NBS-HS-000017, Rev 00, Las Vegas, Nev., 2000.
Fabryka-Martin, J. T., S. J. Wightman, B. A. Robinson, and E. W. Vestal,
Infiltration processes at Yucca Mountain inferred from chloride and
chlorine-36 distributions, Milestone Rep. 4317, Los Alamos Natl. Lab.,
Los Alamos, N. M., 1994.
Fabryka-Martin, J. T., A. L. Flint, D. S. Sweetkind, A. V. Wolfsberg, S. S.
Levy, G. J. C. Roemer, J. L. Roach, L. E. Wolfsberg, and M. C. Duff,
Evaluation of flow and transport models of Yucca Mountain, based on
chlorine-36 and chloride studies for FY97, Yucca Mt. Proj. Milestone
Rep. SP2224M3, Los Alamos Natl. Lab., Los Alamos, N. M., 1997.
Flint, L. E., Characterization of hydrogeologic units using matrix properties, U.S. Geol. Surv. Water Resour. Invest. Rep., 96-4342, 61 pp., 1998.
Flint, L. E., Characterization of unsaturated zone hydrologic properties and
their influence on lateral diversion in a volcanic tuff at Yucca Mountain,
Nevada, Ph.D. dissertation, 161 pp., Oreg. State Univ., Corvallis, 2002.
Flint, L. E., and A. L. Flint, Distribution of water potential measured with
heat dissipation probes in underground volcanic tuffs, paper presented at
the Geological Society of America Annual Meeting, Reno, Nev., Nov.
2000.
Flint, L. E., A. L. Flint, C. A. Rautman, and J. D. Istok, Physical and
hydrologic properties of rock outcrop samples at Yucca Mountain, Nevada, U.S. Geol. Surv. Open File Rep., 95-280, 70 pp., 1996.
Flint, A. L., L. E. Flint, J. A. Hevesi, D’F. A. Agnese, and C. C. Faunt,
Estimation of recharge and travel-time through the unsaturated zone in
arid climates, in Dynamics of Fluids in Fractured Rock, Geophys.
Monogr. Ser., vol. 122, edited by B. Faybishenko, P. Witherspoon, and
S. Benson, pp. 115 – 128, AGU, Washington, D. C., 2000.
Flint, A. L., L. E. Flint, G. S. Bodvarsson, E. M. Kwicklis, and J. T.
Fabryka-Martin, Evolution of the conceptual model of vadose zone hydrology for Yucca Mountain, J. Hydrol., 247(1 – 2), 1 – 30, 2001a.
Flint, A. L., L. E. Flint, E. M. Kwicklis, G. S. Bodvarsson, and J. T.
Fabryka-Martin, Hydrology of Yucca Mountain, Rev. Geophys., 39(4),
447 – 470, 2001b.
Flint, A. L., G. S. Campbell, K. M. Ellett, and C. Calissendorff, Calibration
and temperature correction of heat dissipation matric potential sensors,
Soil Sci. Soc. Am. J., 66, 1439 – 1445, 2002a.
Flint, A. L., L. E. Flint, E. M. Kwicklis, J. T. Fabryka-Martin, and G. S.
Bodvarsson, Estimating recharge at Yucca Mountain, Nevada, USA:
Comparison of methods, Hydrogeol. J., 10, 180 – 204, 2002b.
Gardner, W. R., Some steady-state solutions of the unsaturated moisture
flow equation with application to evaporation from a water table, Soil
Sci., 86, 228 – 232, 1958.
Heiberger, T. S., Simulating the effects of a capillary barrier using the twodimensional variably saturated flow model SWMS-2D/HYDRUS-2D,
Master’s thesis, 124 pp., Oreg. State Univ., Corvallis, 1996.
Hevesi, J. A., A. L. Flint, and J. D. Istok, Precipitation estimation in
mountainous terrain using multivariate geostatistics, II, Isohyetal maps,
J. Appl. Meteorol., 31(7), 677 – 688, 1992.
Ho, C. K., Assessing alternative conceptual models of fracture flow, in
TOUGH Workshop >95, Proceedings, edited by K. Pruess, Rep. LBL37200, Lawrence Berkeley Natl. Lab., Berkeley, Calif., 1995.
Ho, C. K., and S. W. Webb, Capillary barrier performance in heterogeneous
porous media, Water Resour. Res., 34(4), 603 – 609, 1998.
Istok, J. D., C. A. Rautman, L. E. Flint, and A. L. Flint, Spatial variability in
hydrologic properties of a volcanic tuff, Ground Water, 32, 751 – 760,
1994.
Klavetter, E. A., and R. R. Peters, Estimation of hydrologic properties of an
unsaturated fractured rock mass, Rep. SAND84-2642, Sandia Natl. Lab.,
Albuquerque, N. M., 1986.
Kwicklis, E. M., and R. W. Healy, Numerical investigation of steady liquid
water flow in a variably saturated fracture network, Water Resour. Res.,
29(12), 4091 – 4102, 1993.
Kwicklis, E. M., A. L. Flint, and R. W. Healy, Simulation of flow in the
unsaturated zone beneath Pagany Wash, Yucca Mountain, in Proceedings
of the 5th International High-Level Radioactive Waste Manage Conference, May 22 – 26, Las Vegas, Nevada, pp. 2341 – 2351, Am. Nucl. Soc.,
La Grange Park, Ill., 1994.
Montazer, P., and W. E. Wilson, Conceptual hydrologic model of flow in
the unsaturated zone, Yucca Mountain, Nevada, U.S. Geol. Surv. Water
Resour. Invest. Rep., 84-4345, 55 pp., 1984.
Moyer, T. C., J. K. Geslin, and L. E. Flint, Stratigraphic relations and
hydrologic properties of the Paintbrush Tuff nonwelded (PTn) hydrologic
unit, Yucca Mountain, Nevada, U.S. Geol. Surv. Open File Rep., 95-397,
151 pp., 1996.
Peters, R. R., and E. A. Klavetter, A continuum model for water movement
in an unsaturated fractured rock mass, Water Resour. Res., 24(3), 416 –
430, 1988.
Rautman, C. A., and A. L. Flint, Deterministic processes and stochastic
modeling, in Proceedings of the 2nd Annual International High-Level
Radioactive-Waste Management Conference, pp. 1617 – 1624, Am. Nucl.
Soc., La Grange Park, Ill., 1992.
Ritcey, A. C., and Y. S. Wu, Evaluation of the effect of future climate
change on the distribution and movement of moisture in the unsaturated
zone at Yucca Mountain, Nevada, J. Contam. Hydrol., 38, 257 – 279,
1999.
Rockhold, M. L., B. Sagar, and M. P. Connelly, Multidimensional modeling of unsaturated flow in the vicinity of exploratory shafts and
fault zones at Yucca Mountain, Nevada, in First International High
Level Radioactive Waste Management Conference, Las Vegas, Nevada,
Proceedings, pp. 153 – 162, Am. Nucl. Soc., La Grange Park, Ill.,
1990.
Roseboom, E. H., Jr., Disposal of high-level nuclear waste above the water
table in arid regions, Geol. Surv. Circ., 903, 21 pp., 1983.
Ross, B., The diversion capacity of capillary barriers, Water Resour. Res.,
26(10), 2625 – 2629, 1990.
Rulon, J. J., G. S. Bodvarsson, and P. Montazer, Preliminary numerical
simulations of groundwater flow in the unsaturated zone, Yucca Mountain, Nevada, Rep. LBL 20553, 91 pp., Lawrence Berkeley Natl. Lab.,
Berkeley, Calif., 1986.
Scott, R. B., R. W. Spengler, S. Diehl, A. R. Lappin, and M. P. Chornack, Geologic character of tuffs in the unsaturated zone at Yucca
Mountain, southern Nevada, in Role of the Unsaturated Zone in Radioactive and Hazardous Waste Disposal, edited by J. W. Mercer, P. S. C.
Rao, and I. W. Marine, pp. 289 – 335, Ann Arbor Sci., Ann Arbor,
Mich., 1983.
Selker, J. S., C. K. Keller, and J. R. McCord, Vadose Zone Processes, 339
pp., CRC Press, Boca Raton, Fla., 1999.
Sinnock, S., Y. T. Lin, and J. P. Brannen, Preliminary bounds on the
expected post-closure performance of the Yucca Mountain repository site, southern Nevada, J. Geophys. Res., 92(B8), 7820 – 7842,
1987.
Sonnenthal, E. L., and G. S. Bodvarsson, Constraints on the hydrology of
the unsaturated zone at Yucca Mountain, Nevada, from three-dimensional models of chloride and strontium geochemistry, J. Contam. Hydrol., 38(1 – 3), 107 – 157, 1999.
Steenhuis, T. S., J. Y. Parlange, and K.-J. S. Kung, Comment on ‘‘The
diversion capacity of capillary barriers’’ by Benjamin Ross, Water Resour. Res., 27(8), 2155 – 2156, 1991.
FLINT ET AL.: INFLUENCE OF VOLCANIC STRATA ON DIVERSION
van Genuchten, M. T., A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils, Soil Sci. Soc. Am. J., 44, 892 – 898,
1980.
Walter, M. T., J.-S. Kim, T. S. Steenhuis, J.-Y. Parlange, A. Heilig, R. D.
Braddock, J. S. Selker, and J. Boll, Funneled flow mechanisms in a
sloping layered soil: Laboratory investigation, Water Resour. Res.,
36(4), 841 – 849, 2000.
Wilson, M. L., Lateral diversion in the PTn unit: Capillary-barrier analysis,
in Sixth International High Level Radioactive Waste Management Conference, Las Vegas, Nevada, Proceedings, Am. Nucl. Soc., La Grange
Park, Ill., 1996.
SBH
4 - 17
Winograd, I. J., Radioactive waste disposal in thick unsaturated zones,
Science, 212(4502), 1457 – 1464, 1981.
A. L. Flint and L. E. Flint, Water Resources Division, U.S. Geological
Survey, Placer Hall, 6000 J Street, Sacramento, CA 95819-6129, USA.
(lflint@usgs.gov)
J. S. Selker, Department of Bioresource Engineering, University of
Oregon, Corvallis, OR 97330, USA.